AI culling
AI photo culling software
Five tools that pick the keepers, and the one that is not AI at all.
AI culling
Five tools that pick the keepers, and the one that is not AI at all.
Culling is the first hour of every job and the least creative. A thousand frames arrive and most of them are duplicates, blinks or missed focus.
AI culling tools sort that pile. They group near-identical frames, check eyes and focus, and propose the keeper from each group.
| Tool | How it chooses | Price as published |
|---|---|---|
| Aftershoot Select | On your own machine. Learns which frames you keep. | Subscription, see vendor page |
| Narrative Select | Built for review. Large previews, grouping, eye and focus checks. | 10, 20, 40 and 60 US dollars a month |
| FilterPixel | Genre models. You pick wedding, conference, sport or portrait before it runs. | 14.99 US dollars a month billed annually, 34.99 for the full workflow, extra shoot 7.99 |
| Optyx | Scores focus and faces so you can filter on technical quality. | Free tier, Pro 6.99 US dollars a month billed annually, 9.99 monthly |
| Excire Foto | Searches a whole library by face and content, not just one shoot. | One-off licence |
| Photo Mechanic | Not AI. The speed benchmark that culling tools are measured against. | One-off licence |
Prices and features on this page were read from each vendor's own website on 25 September 2026. Vendors change plans often. Check the current page before you buy.
Select is the culling half of Aftershoot. It groups near-identical frames, checks eyes and focus, and proposes one keeper from each group.
It learns from what you override, so the second shoot is closer to your taste than the first.
Everything happens on your machine, which keeps client files local and makes the speed a function of your hardware.
Narrative is built for the review itself rather than for automation. Large previews, grouping, and a close-up that shows whether the eyes are sharp.
The AI assists, you decide, and the interface is designed so that deciding is fast.
Plans run from ten to sixty US dollars a month, which puts the entry point below most of the alternatives.
FilterPixel asks what you shot before it starts: wedding, portrait, conference, sport or concert. It then loads a model trained for that genre.
That matters because a keeper at a conference is not a keeper at a wedding. One rewards a clean speaker frame, the other rewards a reaction.
Listed at 14.99 US dollars a month billed annually, with a fuller workflow plan at 34.99 and single extra shoots at 7.99.
Optyx scores frames on focus and faces and lets you filter on the score, which suits a photographer who wants to skim rather than delegate.
There is a free tier, and Pro is listed at 6.99 US dollars a month billed annually or 9.99 billed monthly.
It is the cheapest way to find out whether automated culling suits you at all.
Excire searches a whole library rather than a single shoot. Find every frame with a particular face, or every frame with a given subject, across years of work.
That is an archive tool more than a culling tool, and it earns its place when a client asks for something you shot three years ago.
Photo Mechanic has no AI at all. It is included here because it is the speed benchmark every culling tool is measured against.
It reads embedded previews instantly, so browsing a card of RAW files feels like browsing JPEGs.
Press photographers still use it because a human decision made quickly beats a machine decision made slowly.
This is the real decision. Local tools keep client files on your machine and use your processor. Cloud tools upload the shoot and work anywhere.
If your client contract says images stay in the United Kingdom or the European Union, that rules out most cloud options. There is more on this in client RAW files and cloud tools.
Mariya's verdict. Culling is the safest place to let software decide. A wrong keeper costs you one frame. A wrong retouch costs you the client's face.
Take a shoot you already culled by hand. Run the tool on the same folder. Compare its selection against yours.
Count two numbers. How many of your keepers it found, and how many frames it kept that you would have deleted. The second number is the one that wastes your time later.
A tool reads sharpness, exposure and open eyes. It cannot read the moment.
At a wedding the frame that matters is often technically worse. Someone moved, the focus slipped, and the picture is still the one the family prints.
On a fashion shoot the same thing happens with gesture. A hand caught mid-movement carries the look, and the tool marks it as motion blur.
So the workflow that works is two passes. Let the software throw out the obvious rejects. Then look at what is left with your own eyes.
That order saves the hour and keeps the judgement where it belongs.
A conference: eleven hundred frames across a day, delivery promised for the next morning.
Automated culling reduces that to about two hundred and forty candidates in minutes.
Review takes twenty minutes, and it is spent on the rejects as much as on the keepers.
Two frames come back from the reject pile: a speaker mid-gesture that was marked as motion blur, and a delegate laughing with eyes half closed.
Both end up in the client's selection. That is the argument for reviewing the pile the software throws away.
| Situation | Tool | Why |
|---|---|---|
| High volume, local processing | Aftershoot Select | Runs on your machine and learns your picks |
| Careful review of a small set | Narrative Select | Built for looking, not for automating |
| Events, sport, conferences | FilterPixel | Genre models judge a keeper differently |
| Trying automation cheaply | Optyx | Free tier and the lowest paid plan |
| Finding an old frame in an archive | Excire Foto | Searches a library, not a shoot |
It measures things it can measure. Sharpness at the eye, whether the eye is open, exposure, and how similar one frame is to the one beside it.
Similar frames are grouped, and the group is ranked on those measures. The top of each group becomes the suggested keeper.
Nothing in that list is about meaning. The model cannot see that the blurred frame is the one where the bride laughed.
This is why the reject pile is worth a fast pass. Everything the model got wrong is in there, and it is wrong in a predictable direction.
It is also why culling is the safest job to automate. A wrong keeper costs one frame. A wrong retouch costs the client's face.
Culling is the first hour after a job and the one most photographers put off.
Automated culling moves that hour to fifteen minutes of review, which changes the delivery date more than any editing tool does.
The sensible order is: back up the cards, run the cull, review the rejects quickly, then edit.
Review the rejects, not just the keepers. That is where a tool loses the frame that carried the moment.
Does it run on my machine or in the cloud.
Does it read my camera's RAW files, including a body released this year.
Can it write ratings and colour labels back into Lightroom or Capture One.
How many shoots does the plan cover, and what does an extra one cost.
Can I train it on my own selections, and how many shoots before it helps.
Every price and feature on this page was read from the vendor's own website on 25 September 2026.
Plans change often, and a tool that was a one-off purchase last year may be a subscription now.
Check the current page before you buy, and treat the figures here as a starting point for comparison rather than a quote.
Nothing on this page is sponsored. No vendor paid for a place in the tables, and there are no affiliate links.
This page is written by Mariya Harley, a fashion, portrait and hotel photographer working from a studio in west London.
The tools described here are the ones that reach a working photographer's desk, and the failures described are the ones that reach a client's screen.
Prices and features are read from vendor pages rather than repeated from other articles. The date of that check is printed above each table.
Nothing here is sponsored, and there are no affiliate links. A tool appears because it does the job, or because people ask about it.
What is deliberately absent: benchmark scores copied from press releases, rankings that change with the season, and any claim that software can replace direction, light or responsibility for the result.
If a price here is out of date, the vendor page wins. Tell me and I will correct it.